TORSION DESIGN OF

PRESTRESSED CONCRETE
Paul Zia
Professor and Associate Head
of Civil Engineering
North Carolina State University
Raleigh, North Carolina

W. Denis McGee
Assistant Professor of
Civil Engineering
Washington State University
Pullman, Washington

Design provisions for torsion in reinforced concrete

members have been developed and included in the ACI
318-71 Building Code. These provisions are not
applicable to prestressed concrete.
Based on a thorough review of available research data
on torsion in prestressed concrete, this paper proposes
an extension of the ACI design procedures to cover
torsion in prestressed concrete. The basic approach
consists of determining the concrete contribution to the
ultimate strength of a member and then proportioning
reinforcement for the remaining portion of the
required ultimate strength.
Formulas for the nominal torsional stress and the basic
equation for torsional strength of reinforced members
are presented, and followed by a discussion of the
torsion-shear interaction relation, taking into account
the effect of the prestressing force. The requirements for
both web and longitudinal reinforcements for torsion
are explained.
The proposed procedures are illustrated by a design
example of a precast prestressed L-girder in a roof
framing subjected to combined torsion, bending, and
shear. It is believed that the proposed procedure is
reasonably conservative and future refinements and
simplifications are possible when more complete
research data become available.

46
 The problem of torsion in concrete
structures has received considerable at-
tention of design engineers in recent
years. In 1971, the ACI Budding Code
(ACI 318-71) 1 included for the first
time explicit requirements of torsion de-
sign for reinforced concrete developed
on the basis of substantial research da-
ta. However, the code provisions are
not applicable to prestressed concrete
members, for which research has been
in progress but design criteria have yet
to be developed.
This paper presents a proposed de-
sign procedure for torsion in prestressed
concrete, based on a thorough analysis Paul Zia
of 394 test results available in the liter-
ature. 2 The proposal is in close parallel
with the ACI Code procedure for rein-
forced concrete.
The basic approach consists of deter-
mining the concrete contribution to the
ultimate strength of a member and then
proportioning reinforcement for the re-
maining portion of the required ulti-
mate strength. It is believed that the
proposed procedure is reasonably con-
servative and future refinements and
simplifications are possible when more
complete research data become avail-
able.
In developing the proposed proce-
dure, the authors have benefited from W. Denis McGee
helpful discussions with several mem-
bers of the ACI Committee 438 on Tor-
sion. However, this proposal should not
be regarded as a committee-endorsed
document.
torsional stress of a rectangular section
In the following discussion, the usual can be expressed by:
capacity reduction factor 0 is not in-
cluded for the sake of convenience. T.
Tu— m, 2y (1)

e where
TORSIONAL SHEAR STRESS x = shorter side of rectangular sec-
tion
y = longer side of rectangular sec-
Rectangular section
tion
A prestressed concrete member with- a = torsion coefficient
out web reinforcement fails, under tor- It will be recalled that Eq. (1) is of
sion, in the form of skew bending. The the basic form for either the elastic the-

PCI Journal/March-April 1974 47

 0.55

plastic
05 1 coeff.
k i

6 0.45

00.40 ^• ._
o !:
V

0 035 ; j' _ ,-AC I of 1/3

° 030 ; 0.75+x/y

elastic
0.25 coeff.

0.20 position
of ave.f

0.150
1 2 3 4 5
aspect ratio y/x
Fig. 1. Variation of torsion coefficients with aspect ratio.

ory, the plastic theory or the skew where

bending theory of torsion. The coeffi- f,. = modulus of rupture
cient a varies from 0.208 to 1/3 in the a- = average prestress on section
elastic theory and from 1/a to 1/a in the The factor (0.85 f,.) is nearly equal to
plastic theory. For simplicity, the ACI the tensile strength of concrete f,' and
Code adopted a value of 1/3 in accor- may be taken as 6Vf j . Hence, corn-
dance with the skew bending theory. 3 bining Eqs. (1) and (2):
Hsu4 has shown that the torsional
capacity of a prestressed section with- Tu = 6Vf^ ^1 + 10 a-/f0 — IXx2y
out reinforcement, or the cracking
torque of a prestressed section with re- or
inforcement is reached when: _ _ 3
a
Tu = (0.85 fr) 10 a- /f0' (2) x2y 6 Vfc lrl + 10 a-/f0 ( )

48
 Using the test results of 218 concen-
trically or eccentrically prestressed rec-
tangular beams, the value of a has been
computed for each beam by Eq. (3)
and these values are plotted against the
aspect ratio y/x in Fig. 1. Also plotted
for comparison in Fig. 1 are the curves
for the elastic and plastic torsion coeffi-
cients as well as the value currently
specified by the ACI Code.
It can be seen that for sections of low
aspect ratio, the ACI 318-71 value of
1/. could be quite unconservative.
Since, as will be shown later, the con-
tribution of the concrete to the ultimate
torsional strength is a much larger por-
tion of the cracking strength for pre-
stressed members than for non-pre-
stressed members, it is prudent and es-
sential that the nominal torsional stress
be evaluated more accurately. There-
fore, it is proposed that the coefficient
a in Eq. (1) be taken as:
0.35
« = 0.75 + x/ y (4)
which represents a reasonably conserva-
tive lower bound to the experimental
data.

Table 1. Comparison of experimental cracking torque of prestressed flanged sec-

tions with Eq. (5).

Inves- Beam E(ax2y)

Tex Tth* Tex
tigator No. (in.3) (in.-k) (in.-k) Tth

Wyss, et. al. Al 294 253 234 1.08

(I-beams)ḏ(5) A2 294 256 234 1.10
A3 294 300 233 1.29
A4 294 316 233 1.35
A5 294 329 233 1.42
A6 294 313 233 1.35
B1 294 203 185 1.10
B2 294 216 185 1.17
B3 294 222 188 1.19
B4 294 222 188 1.19
B5 294 209 188 1.12
B6 294 200 188 1.06
Ziaḏ (6) 0.25T1 52.9ḏ
(41.5)** 51.12 40.3 (31.6)** 1.27 (1.62)**
(T-beams) 0.25T2 52.9ḏ
(41.5) 51.98 40.3 (31.6) 1.29ḏ(1.64)
2.25T1 52.9ḏ
(41.5) 35.20 39.9ḏ(31.3) 0.89ḏ (1.13)
2.25T2 52.9ḏ
(41.5) 32.92 39.9ḏ(31.3) 0.83ḏ (1.06)
2.75T1 52.9ḏ
(41.5) 33.96 39.8 (31.2) 0.88 (1.12)
2.75T2 52.9ḏ
(41.5) 33.16 39.9ḏ(31.3) 0.83 (1.06)
Zia (6) 0.75I1 57.1 (49.4) 28.0 39.4 (34.1) 0.71 (0.82)
(I-beams) 0.75I2 57.1 (49.4) 35.0 39.4 (34.1) 0.89ḏ (1.03)
3.I1 57.1 (49.4) 30.0 39.7 (34.4) 0.75ḏ (0.87)
3.I2 57.1 (49.4) 33.0 39.7ḏ(34.4) 0.83 (0.96)
3.5I1 57.1 (49.4) 48.5 41.2ḏ(35.6) 1.18 (1.36)
3.512 57.1 (49.4) 46.0 41.2ḏ(35.6) 1.11 (1.29)

*Computed from Equation (5).

**Neglecting outstanding flanges.
+Cracking torques for these members were not available; T was
taken as the elastic limit, which is less than the actuaTTX
cracking load.

PCI Journal/March-April 1974 49

Table 2. Comparison of experimental cracking torque of prestressed box sections
with theory.

Beam x, y Equiv. T T * Tex

Investigator Core square 4h/x ex th
No. (in.) core (in. -k) (in.-k) Tth

T 9 5" x 5" 5" x 5" 0.89 127.2 113.9 1.12

Johnston H-0-0-1 12 8.5" 7.5" x 7.5" 0.75 120 99.5 1.21

and Zia (9)
H-0-3-1 12 8.5" 7.5" x 7.5" 0.75 120 73.7 1.63

H-0-6-1 12 8:5"c 7.5" x 7.5" 0.75 115 73.7 1.56

*Computed from Equation (1) multiplied by 4h/x.

Flanged section Box section

For a flanged section, the same as-
sumption implied by ACI 318-71 will
be used. That is, the torsional strength
of a flanged section can be expressed as
the sum of the strengths of the individ-
ual components. Hence:
Tu =^ 2y
Eq. (5) is identical to the ACI 318-
71 Eq. (11-16) except for the modifi-
cation of torsional coefficient a as ex-
pressed by Eq. (4). The section may
be divided into component rectangles
such that the quantity l.ax2y would be
a maximum.
Comparison of Eq. (5) with test re-
sults reported in the literature indicates
that the method is reasonably safe as
shown in Table 1. However, it should
be cautioned that only two investiga-
tions5 ' 6 have been reported, consisting.
of 24 test specimens, all being some-
what stocky in cross section. The mar-
gin of safety seems to reduce as the
cross section becomes more stocky such
as the specimens tested by Zia. For
such disproportionate sections, it would
seem logical to disregard the small out-
standing flanges.
Based on test results, ACI 318-71
specifies that for reinforced concrete
box sections with a wall thickness h not
less than x/4, where x is the overall
width of the box section, the torsional
strength of the box section may be
taken as that of a comparable solid rec-
^ Jr) tangular section with the same overall
dimension.
If the wall thickness h is less than
x/4 but greater than x/10, the torsion-
al strength of the box section is reduced
and the strength reduction is accounted
for by a reduction factor 4h /x multi-
plied to the denominator on the right-
hand side of Eq. (1).
This design approach seems also ap-
plicable to prestressed concrete box sec-
tions. Although only four tests of pre-
stressed concrete box sections under
pure torsion have been reported in the
literature that can be used for compari-
son, application of the ACI design rule
for these test results indicates a consid-
erable margin of safety in all cases as
shown in Table 2.
For box sections with a very thin
wall, h approaching x/10, caution
should be exercised against possible

50
wall buckling or crushing, particularly
when high prestress is applied.
BASIC EQUATION FOR
TORSIONAL STRENGTH
The basic equation for torsional
strength has been developed from the
test results of prestressed beams under
pure torsion. Tests conducted by Hsu7
show that the ultimate torsional
strength can be expressed as the sum of
the strength contributed by the con-
crete and the strength contributed by
the web reinforcement, for both pre-
stressed and non-prestressed beams
(see Fig. 2). The effect of prestress is
to increase the contribution of the con-
crete to the ultimate torsional strength,
while the contribution of the reinforce-
ment remains unchanged.
Referring to Fig. 3, it can be seen
that for non-prestressed beams the con-
tribution by concrete To is only a por-
tion of the torsional strength Tu5 of a
corresponding plain concrete section.
Likewise, for prestressed beams, the
contribution by concrete To is also a
portion, but a greater portion, of the
torsional strength of a corresponding
prestressed concrete section without
web reinforcement. Accordingly, the ba-
sic strength equation for prestressed

600ҟ
10"
DO
15"
500

400ҟ
0 ^^j

300ҟ
^

200ҟ
6\/``' o prestressed
o nonprestressed
••e

00 100 200 300 400 500

x^ yI As fsy /s (ink)

Fig. 2. Tests on prestressed and nonprestressed reinforced concrete beams (after

Hsu).

PCI Journal/March-April 1974 51



Tu ^a that of the web reinforcement, then the

parameter m should be replaced by
t^^Jn mfj /f3i, in which f ly is the yield
strength of the longitudinal reinforce-
ment. Hsu has recommended that:
0.7 mfl5 /f8 1.5 (7)
and

7
yl/xl 2.6 (8)
In Fig. 3, the cracking (or ultimate)
lb 0 torque of a plain concrete member can
be obtained by substituting 6Vf,,' for
Tu in Eq. (1). Thus:

I—ҟ
°i T., = ( ax2y)6 V/ff (9)
^ Amin. reinf. The ordinate T° has been deter-
mined by Hsu experimentally as:

T0 =0.4(3xZy)6V
x i y1 A s fsy /s
= 0.133(x2y)6Vf0'
Fig. 3. Relation between ultimate Hence:
torque for prestressed and non pre- To =T,,, /1 +10a-,/f 0' — (T„ P — To)
stressed reinforced, rectangular beams = Tu5 [V I + 10 a-/f,, -
(after Hsu). (1 — T0/T)]
_ (ax2y)6 Vi (V 1 + 10 o-/f0' — k)
(10)
where
k=1—T0/T,,5= 1-0.133/a (11)
Thus, by combining Eqs. (6) and
beams can be expressed as: (10), the basic strength equation be-
comes:
1 y 1 A cf8y (6)
T u = To + H x
s 1',, = ( a x`y)6 Nlf (V 1 + 10 a-/f,,' — k) +
and 1 y 1 A tf3v (12)
fl = 0.66 m + 0.33 y1/x1 s
where Likewise, for a box section:
To = strength contributed by con- T,, = [(ax2y)6 Vf0' X
crete
xl = shorter leg of a closed stirrup
(V1 +10a-/f0 - k) +
yl = long leg of a closed stirrup
At = cross-sectional area of one leg cX SYSAtfSY]
of a stirrup (13)
s
f8y = yield strength of stirrup
and, for a flanged section:
s = spacing of stirrup
m = volume ratio of longitudinal 1 ',, = ( ax2 y)6 Vf x
reinforcement to web rein- (Vi + 10a-/f^ - k) +
forcement 1 y1Atf8 v (14)
If the yield strength of the longi- s
tudinal reinforcement is different from in which the value of k is determined

52
Table 3. Comparison of experimental ultimate torque of concentrically prestressed
rectangular beams with Eq. (12).

Inves- Seam Inves- Beam T,

Tex Tth* Tex Tth* Teic
tigator No. (in.-k) (in.-k) Tth tigator No. (in.-k) (in.-k) Tth

Mukherjee and
Chandler, C1 59.32 54.67 1.09 106 140.8 97.05 1.45
Warwaruk (12) 206 146.4 121.24
et. a1. 03 79.00 62.07 1.27
C5 75.60 66.94 1.23
(10) C9 85.10 72.47 1.17
94.50 78.86 1.20 Okada, et. al. RbIII-1I 47.0 25.05 1.88
Cli
106.40 (13) RbIII-21 47.6 25.05 1.90
C15 85.64 1.24
C7 54.77 58.68 0.93
C13 59.10 49.18 1.20
41.08 42.54 0.97 Superfeskyḏ
(14) SP17 213.0 141.19 1.51
C17
C19 51.60 62.00 0.83 SP18 246.0 229.24 1.07
SP1261 164.90 173.74 0.95 SP19 245.0 188.70 1.30
SP1262 134.65 130.15 1.03 SP20 215.0 169.94 1.27
SP1269 151.15 155.25 0.97 4S1 63.5 35.63 1.78.
SP1270 129.90 112.80 1.15 452 77.5 58.68 1.32
SP1241 82.90 67.71 1.22 453 92.7 61.50 1.51
SP1242 81.40 67.71 1.20 4S4 83.1 63.78 1.30
SPC1243 70.95 58.51 1.21 455 95.8 68.91 1.39
SPC1245 103.00 84.67 1.22 4S6 72.0 42.30 1.70
SP1247 93.70 77.99 1.20 457 80.6 65.75 1.23
SPC941 52.20 46.05 1.13 4S8 88.3 67.85 1.30
SP1263 175.15 172.89 1.01 4S9 99.4 71.14 1.40'
SP1264 136.15 129.31 1.05 4S10 101.3 75.80 1.34
SP1271 162.40 155.39 1.05 4511 49.0 40.00 1.23
SP1272 121.15 112.94 1.07 4S12 88.7 63.90 1.39'
SPI 152.15 118.55 1.28 4S13 90.7 66.02 1.37
SPII 142.65 118.58 1.20 4S14 89.8 68.64 1.31.
SP1267 169.25 170.91 0.99 4515 94.7 72.82 1.30
1
Ziaḏ
(6) ORW1 50.64 40.25 1.26
0RW2 57.44 40.25 1.43
Mukherjee SP1 190.8 129.79 1.47
Kempḏ (11) SP2 171.0 122.79 1.39
SP3 233.8 204.82 1.14 GangaRao 1-0 79.86 81.76 0.98
SP4 198.6 164.45 1.21 and Zia (15) 2-0 111.68 89.50 1.25
SP5 200.8 153.85 1.31 2-C 136.20 89.07 1.53
SP6 153.0 122.99 1.24 3-0 104.70 104.29 1.00
SP7 156.0 110.51 1.41 4-0 120.00 102.38 1.17
SP8 186.0 190.83 0.97
SP9 185.0 155.62 1.19 Hsu (7) P1 382 318.76 1.20
SP10 178.0 143.42 1.24
SP11 207.0 163.39 1.27 P2 440 384.67 1.14
SP12 205.0 153.93 1.33 P3 500 468.25 1.07
SP13 232.0 243.96 0.95 P4 534 563.47 0.95
SP14 232.0 200.34 1.16 P5 424 349.29 1.21
SP15 203.0 184.46 1.10 P6 496 414.17 1.20
P7 530 504.05 1.05
P8 546 687.99 0.79
Jacobsen(16) 306 177.9 122.09 1.46
*Computed from Equation (12)

on the basis of the largest component TORSION-SHEAR INTERACTION

rectangle where the web reinforcement
is usually placed.
By comparison with available test re-
sults, it has been shown that, with only
a few exceptions, the above three equa-
tions are reasonably and consistently on
the safe side (see Tables 3 through 5).2
It would appear, therefore, that these
equations are acceptable as the basis of
design.
Tests2 have shown that the interac-
tions between torsion and shear for pre-
stressed beams without web reinforce-
ment can be adequately represented by
a circular curve. By following the same
development adopted by ACI 318-71,
it can be shown that the permissible
shear stress for torsion and shear are,
respectively:

PCI Journal/Marcia-April 1974 53

Table 4. Comparison of experimental ultimate torque of eccentrically prestressed
rectangular beams with Eq. (12).

Ines- Beam T Tḏ * T
ex th ex
tigator No. (in.-k) (in.-k) Tth

Chandler, et. E2 52.82 54.90 0.96

al.ḏ (10) E4 64.70 62.94 1.03
E6 70.20 66.94 1.05
E10 79.70 72.47 1.10
E12 92.20 78.86 1.17
E16 107.82 85.64 1.26
E8 50.44 58.68 0.86
E14 61.90 49.18 1.26
E18 41.35 42.54 0.97
E20 53.10 62.00 0.87
SPE1244 66.70 58.51 1.14
SPE1246 97.00 84.67 1.15
SPE942 44.70 46.05 0.97

Ganga Rao (15) 5-0 106.80 109.30 0.98

and Zia 6-0 119.50 107.96 1.11

Jacobsenḏ
(16) 326 154.2 143.50 1.07

Mukherjee and 126 147.5 99.50 1.48

Warwaruk (12) 226 152.3 121.23 1.26

(6)
Ziaḏ 2RW1 48.44 40.33 1.20
2RW2 52.24 40.09 1.30
2.5RW1 53.44 39.81 1.34
2.5RW2 52.44 39.81 1.32

McGee and W2e-4.5-LL 105 63.99 1.64

Ziaḏ (2)

Woodhead and I1-1 105 70.89 1.48

McMullenḏ (17) 111-4 125 87.50 1.43

*Computed from Equation (12).

6 1 k)
To _ where
TC _
1 + (^ v5 1 2(15)
2 T0' = ^/f' (^/ + 10 a-/f0'
-/f0'- (17)
L v; = lesser of vj and v defined
h+ T
by ACI Eqs. (11-11) and 11-12), re-

V a=
v 0'
(,a r, )2 (16)
spectively.
Q- 2 (Tc, 1 (18)
1+ Uu J
54
 It is noted that r6' and v 0 ' in Eqs. taking into account the torsion-shear
(15) and (16) are the permissible shear interaction, we have:
stresses for torsion and flexural shear, SZxiyiAtfB„
respectively, if torsion or shear is act- T. = Y-ax 2 yr 0 +
s
ing alone. The factor in the denomi- _ Y.ax2y-r^
nator of these equations accounts for
Transposing terms:
the interaction between torsion and
shear. For a discussion of Eq. (18), see SLx1 y1 Aa f sy
'x2y(?u -r0)=
Reference 2. S

or
WEB REINFORCEMENT At_ (ru-Tc)s laxly
(19)
FOR TORSION 12xiyif"
in which, for simplicity, the coefficient
Recasting the basic equation for tor- fZ may be taken as:
sional strength discussed above and 11= 0.66 + 0.33 y1/x1

Table 5. Comparison of experimental ultimate torque of flanged and box sections

with Eq. (14) and Eq. (13), respectively.

Inves- Beam T th
T*
ex Tex
tigator No. (in. -k) (in. -k) Tth

Wyss,ḏ al.
et.ḏ A2 318.0 194.65 1.63
(5) A3 391.0 223.4 1.75
A4 430.0 255.9 1.68
A5 477.0 313.7 1.52
A6 493.0 385.9 1.28
B2 219.0 137.2 1.60
B3 296.0 163.4 1.81
B4 320.0 195.3 1.64
B5 544.0 252.9 2.15
B6 582.0 325.3 1.79

Zia (6) 0.25TW1 50.20 34.42 1.46

(T-beams) 0.25TW2 54.44 34.42 1.58
2.25TW1 41.00 33.79 1.21
2.25TW2 46.70 33.79 1.38
2.75TW1 46.84 33.59 1.40
2.75TW2 50.56 33.59 1.51

0.75IW1 54.00 31.13 1.74

Zia (6) 1.61
0.751W2 50.20 31.13
(I-beams) 1.94
3IW1 59.72 30.77
3IW2 57.76 30.69 1.88
3.5IW1 59.00 30.58 1.93
3.5IW2 62.10 30.58 2.00

Johnston H-0-3-1 210 109 1.93

and Zia (9) H-0-6-1 176 74 2.38

*Computed from Equation (14) or Equation (13).

PCI Journal/March-April 1974 55

 Thus, the web reinforcement is re-
quired for the torsional shear stress in
excess of that carried by the concrete.
The web reinforcement for torsion must
be in the form of closed stirrups and it
should be added to the web reinforce-
ment for flexural shear. Note that the
latter could be open stirrups.
MINIMUM AND MAXIMUM
WEB REINFORCEMENT
To avoid brittle failure, a minimum
amount of web reinforcement must be
provided to resist both torsion and
shear. A series of tests 2 with prestressed
rectangular beams, containing a mini-
mum amount of web reinforcement as
specified by ACI 318-71 for shear, have
shown that the post-cracking ductility
of the beams was insufficient under
high ratios of torsion and shear. In view
of the sudden and explosive nature of
the torsional failure, and the lack of
sufficient research data, it is suggested
that a beam should be reinforced for no
less than its cracking torque.
At the other extreme, considerations
should be given to the possible danger
of over-reinforcing a member such that
a compressive failure of the concrete
might occur before the reinforcement
yields. To avoid this type of failure, an
upper limit of web reinforcement
should be established by specifying a
maximum permissible nominal torsional
stress. If the design stress exceeds this
limit, then a larger beam section would
have to be used.
Since the prestressing force exerts a
compressive load on a member before
it is subjected to torque, it seems logical
that the upper limit on the nominal tor-
sional stress should be a function of the
prestress ratio o- /f'0. Based on this rea-
soning, it will be assumed that the max-
imum torque a rectangular section can
carry is:
56
T(max) = 2yC /f,' 1/1 + 10 v/f
(20)
in which C is an unknown coefficient
yet to be determined. It is assumed
that C is a function of a-/f0'.
For the extreme case of o-/f,' = 0,
i.e., reinforced concrete members with-
out prestress, ACI 318-71 specifies 12
Vf^ as the limiting stress. Thus:
T.(ma,) = 3 x2iy 12 Vfc = O x2y C Vfc
or C = 4/a.
For a typical cross section with an
aspect ratio of two:
_ 0.35 0.35 _
a- 0.75-1-0.5 1.25-0.28
and
C = 4/0.28 --x= 14
Hence, in Eq. (20), the coefficient C
can be taken as 14 for o-/f0' = 0.
For members subjected to high val-
ues of a-/f0', Zia° has shown that a
transition from tension failure to com-
pression failure would take place at
a-/f0' 0.6. Since the tensile strength
of the concrete can be taken as 6V f 0' ,
it is then reasonable to limit the maxi-
mum nominal torsional stress as 6Vf0'
for a-/f0' = 0.6. Assuming a linear func-
tion of a-/f0' between these two limit-
ing values for C, one obtains:
C = 14 — 13.33 (o-/f 0')Ɂ (21)
In Fig. 4, test results are compared
with Eq. (21) and it is noted that, with
a few exceptions, all the data represent-
ing tension failure fall below the curve
of Eq. (21) and the data for those
beams retorted to have failed in com-
pression fall above the curve. It would
seem that the coefficient C as repre-
sented by Eq. (21) is a very reasonable
value to be used as the upper limit of
the nominal shear stress for both rec-
tangular and flanged sections.
Eq. (20) can be extended to the case
of combined torsion and shear, again
 tens. comp.
fail. fail.
20 Rect. section
conc. p/s o 0
eccen. p/s •
18 Flanged sec. 0 0

16

14
C °
12 °^: o
0 °
0 (9 ° •g o°o o0
10
o
•
W 0 ° s ^o

8 • o 0 0
ep • •

G ° C =14 -13.33 a-/f'

0
0 0.2 0.4 0.6
0 / fc

Fig. 4. Comparison of Eq. (21) with tests results for upper limit of nominal torsion
stress.

based on the circular interaction rela- where

tion. Thus:
C'= C1/1+10a-/fo
C,^fc
^'u(mam) = 2 (22) C = 14 — 13.33
\ 1O
1 + [( c Tull
\ 10/1/`f' /J The term 10 f^ in Eq. (23) is based
on the ACI Code upper limit for flex-
)"(max) = (23) ural shear stress and is obtained by tak-
[()()]2
1 + g a = V in Section 11.6.4 of ACI
318-71.2

PCI Journal/March-April 1974 57


LONGITUDINAL TORSION
REINFORCEMENT
To resist the longitudinal component
of the diagonal tension induced by tor-
sion, longitudinal reinforcement must
be provided in addition to that re-
quired by flexure. This longitudinal re-
inforcement should be approximately of
equal volume as that of stirrups for
torsion in order that both will yield at
failure, i.e., m f l5 / f sy = 1. Therefore:
(xi + yi) fsv
A1 l = S to be neglected if the torsional shear
fly
A l is the total area of longitudinal
torsion reinforcement and must be dis-
tributed around the inside perimeter of
the closed stirrups.
The question of whether the pre-
stressing steel can be considered as a
part of longitudinal torsion reinforce-
ment should be considered. The pre-
stressing steel is provided primarily for
resisting flexure and often is not effec-
tively distributed around the perimeter
of the closed stirrups.
Tests2 have shown that without addi-
tional longitudinal reinforcement of
mild steel, prestressed beams would fail
abruptly under high torsion. In light of
this observation, it is recommended that
until more research data become avail-
able, only the prestressing steel in ex-
cess of that required for flexure and lo-
cated around the perimeter of closed
stirrups should be considered as a part
of the longitudinal torsion steel, having
an equivalent area of:
A (mild)eq — Ap s(eff) fys(off)!!l1/
in which APS(eff) is the excess area of
prestressing steel and f,, ( ,ff) is the ef-
fective stress in prestressing steel.
OTHER LIMITATIONS
To ensure the development of ulti-
mate torsional strength and to control
58
cracking and stiffness at service load,
the maximum yield strength of rein-
forcement should be limited to 60,000
psi and the maximum spacing of stir-
rups should not exceed (x l + yl)/4
nor 12 in. If required, longitudinal tor-
sion reinforcement of not less than a
No. 3 bar should be distributed around
the stirrup at no more than 12 in. apart.
In any event, at least one No. 3 bar
should be placed at each corner of the
stirrup.
For reinforced concrete members,
ACI 318-71 permits the torsional effect
(24)
stress is less than 25 percent of the
cracking stress. This same requirement
would seem justified for prestressed
beams. That is to say, the torsional ef-
fect may be neglected in design if the
torsional stress r,,, is less than:
1 .5\/fo' V1+10cr /f0'
SUMMARY OF PROPOSED
DESIGN PROCEDURE
The design procedure proposed here-
in may be summarized as follows:
1. Determine the nominal torsional
stress according to Eq. (5). The stress
Tu should not exceed that specified in
Eq. (22).
2. If T < 1.5 Alf' Vi + 10 cr/f0' ,
torsion may be neglected.
3. If T,.> T. according to Eq. (15),
determine the required web reinforce-
ment from Eq. (19). Check the mini-
mum web reinforcement requirement.
4. Determine the required longitudi-
nal torsion reinforcement according to
Eq. (24). Check the minimum require-
ment.
DESIGN EXAMPLE
To illustrate the above design proce-
dure, let us consider the design of tor-
sional reinforcement in a prestressed
 Inverted tee ledger girder

c^ I a I I I _- Precast conc.
1 p r 8 wide dbl. tee roof columns (typ.)
i M

0 d -
O '"
O LI
0 (Columns support
M floor below -
r L"- girder - (spandrel) industrial loading

30'-0° 30'-0° I 30'-0° 30'-0° 30'-0°

O @
Fig. 5. Partial plan of precast concrete roof.

concrete L-girder in a roof framing sim-
ilar to that given in the PCA Design
Notes18 as shown in Fig. 5. The design
live load plus insulation and roofing is
40 psf. For this loading, a double tee
8DT18A and a L-girder 18LB30 can be
chosen according to the PCI Design
Ilandbook.19
From an analysis for flexure, thirteen
i -in. diameter 7-wire strands (270 ksi
grade) are required for the L-girder
with the eccentricity of prestress at
midspan being 7.3 in. and that at the
support being 3.65 in. Assume that the
loss of prestress is 22 percent; f'. =
5000 psi; f'ci = 3500 psi; fn = fly = 40
ksi. In the following calculations, the
usual capacity reduction factor for
shear and torsion (A = 0.85 will be ap-
plied.

(a) Calculate Tu and V,,, for spandrel

WDL = 30' X 0.062 = 1.85 kips per ft (double tees)
0.45 kips per ft (spandrel beam)
2.30 kips per ft
WLj = 0.040 X 30' = 1.20 kips per ft
Assume no continuity at columns.
DL = 1.4x2.30X30X2=48.3kips

LL = 1.7 x 1.20 X 30 x 2 = 30.6 kips

V,, at centerline of support = 78.9 kips
Neglecting restraints from double tees and walls.

PCI Journal/March-April 1974 59

 DL = 1.4 X 1.85 x (9/12) x 15' = 29.1 ft-kips
LL = 1.7 X 1.20 x (9/12) X 15' = 22.9 ft-kips
Tu at centerline of support = 52.0 ft-kips (or 625 in.-kips)

(b) Calculate shear stress v,^

V,, at "d = 22.7 in." from support = 78.9 X 13/15 = 68.4 kips
= V,/4b 4,d = 68,400/0.85 X 12 x 22.7 = 295 psi

(c) Calculate torsion stress with Eq. (5)

?'u = Tu/4(^.ax2y)
For maximum J.ax 2 y, consider beam web and outstanding ledge as compo-
nent rectangles.
_ 0.35
= 0.304 (beam web)
«I =075+12/30
.
_ 0.35
= 0.280 (ledge)
a2- 0.75+6/12
.ax2y = 0.304 X 122 X 30 + 0.280 X 62 X 12
=1313+121=1434in.3
T,, at "d" from face of support = 625 X 13/15 = 541 in.-kips
= 541/0.85 X 1434 = 445 psi
Effective prestress = 0.78 X 189 = 147 ksi
o- = 13 x 0.153 x 147/432 = 0.677 ksi = 677 psi
u/ f0' = 677/5000 = 0.135
1.5 V/0' /1 + 10 a/ f = 1.5 x 70.5 X 1.53 = 162 psi < 677 psi
Requires torsion design.

(d) Check Tu(maz) and v,j („ a^) with Eqs. (22) and (23)
C=14- 13.33(cr/f 0')= 14 - 13.33(0.135) = 12.2
C'=C-V1+10cr /f0' = 12.2 x 1.53 = 18.7
C' V fcɁ
_ 18.7V5000

(e) Calculate torsion stress and shear stress carried by concrete To and v0
with Eqs. (15) and (16)
To = 6\/fo (V 1 + 10 o-/f' - kweb)
k,,,e b= 1 - 0 .13 3/x1 = 1- 0 .133/0.304 = 0.563
= 6x/5000 (-/1 + 10(0. 135) - 0.563) = 412 psi

60
 vo = lesser of v, ti and v,w from ACI Eqs. (11-11) and (11-12)
= v aw =3.5 1/fo +0.3ff0 +b
VP

w
= 3•5 V5000 + 0.3(677) + 5930/12 x 22.7
= 247 + 203 + 22 = 472 psi
1
R () (412)=0.573
T^ 412

2 1 295\2
1+ (^j Tu) 1+( • 445)
412
_ = 270 psi
x/2.34
ye, 472
vo = _
( T \\
1+ 1 F0.573 445)
472
= 358 psi
— \/L75

(f) Calculate web reinforcement for torsion with Eq. (19)

x1 =12-2X1.5=9in.
y1 =.30-2X1.5=27 in.
11 =0.66+ 0.33 (yl/xl)
= 0.66 + 0.33 (27/9) = 1.65 > 1.5
Use fZ =1.5
— (Tu — Tc)s XicYx2 y
Ae
iyif8v
A t — (445 — 270) x 1434 -
T 1.5x9x27x40,000 0.0173 sq in. per in.

Maximum s =
xl 4 yl
= 9+27 9 in. < 12 in.
4
At =0.0173X9=0.155 sq. in.
Use No. 4 closed stirrups at 9 in. on center

(g) Check minimum web reinforcement for torsion

Minimum web reinforcement should be what is required to develop cracking
torque of section.
Torsional stress atcracking:
Ter = 6Vf" V1 + 10 v/fo
= 6 i/5000 V1 + 10(0.135) = 650 psi
Torsional stress carried by concrete under pure torsion is T0' = 412 psi.
At — (650-412)x 1434
s 1.5 x 9 x 27 x 40,000 = 0.0234 sq in. per in.
At = 0.0234 X 9 = 0.21 sq in. — 0.20 sq in.
Thus, No. 4 closed stirrups at 9 in. on center is adequate.

PCI Journal/March-April 1974 61

(h) Calculate web reinforcement for shear
v, = 295 psi
v, = 358 psi > v,
But v, > v,/2; so minimum web reinforcement is required.
50b,^,s =
50 x 12 x 9 /40,000 = 0.135 sq in.
Av = f8^

A,,/2 = 0.0675 sq in. << 0.20 sq in.

Thus, the minimum web reinforcement provided for torsion is more than
enough for shear.

(i) Calculate longitudinal torsion reinforcement with Eq. (24)

A i = 2(A t/ s) (xi + Yi) (fsv/f 1 5)
= 2(0.0173) (9 + 27) = 1.24 sq in.
Distribute A l around perimeter of stirrups with maximum spacing of 12 in.
Eight bars are required.
Use four No. 4 bars and four No. 3 bars.
Area provided = 4 X 0.20 + 4 x 0.11 = 1.24 sq in. (ok)

Design reinforcement for ledge

Load on ledge:
DL = 1.4 x 1.85 = 2.59 kips per ft
LL = 1.7 X 1.20 = 2.04 kips per ft
V,, = 4.63 kips per ft
d = 12 - 2 = 10 in.
Check for beam action:
v — Vu — 4630 - = 141 psi (ok)
= 45.5 psi < 2^/f^
u— Obwd-0.85X12X10
Check for slab action:
Vu — 4x4630
= 282 psi (ok)
vu = ^b o d 0.85 X21 x10 = 104 psi < 4\/f^
M,u = 4.63 X 3 = 13.89 in.-kips per ft
jd , 9.5 in.
Mu 13890 = 0.0405 sq in. per ft
`^ S — 4 jd fy - 0.9 X 9.5 x 40,000
Even if one-third more than the above steel area is provided, the design is
controlled by ACI Code Section 10.5.2.
A8 = 0.002 X 12 x 10 = 0.24 sq in. per ft
Use No. 4 bars at 10 in. on center and one No. 4 bar in each corner.
Check shear friction:
V _ 4630 - 0.098 sq in. per ft < 0.24 sq in. per ft
AVt — O fy p — 0.85 X 40,000 X 1.4
The reinforcement details for the L-girder are shown in Fig. 6.

62
 12"
ҟ
4 sti r. @ 9' o.c.
tt
with 44&
j 4 tt 3 long.
bars
1
JL
c.g.c.
N
T ® 4412"
c .g.s.
18" 410'o.c.
Fig .6. Reinforcement details f or L-girder.
CONCLUSIONS
A procedure for design of torsion in
prestressed concrete members has been
proposed, which follows the same ap-
proach as ACI 318-71 for design of tor-
sion in reinforced concrete members.
Comparison2 with available test results
of 132 beams under combined loading
as reported in the literature indicates
that the proposed procedure is reason-
ably conservative for rectangular beams
and quite conservative for flanged and
box beams. The application of the de-
sign procedure has been illustrated by
a design example.
When more complete research data
becomes available, especially for pre-
stressed flanged and box members, the
procedure can be further refined and
simplified. In the meantime, design ta-
bles and curves can be easily prepared
to aid the designers in reducing the
amount of computation.
PCI Journal/March-April 1974 63
ACKNOWLEDGMENTS
This paper is based on the results of
one phase of a continuing torsion re-
search program supported by the Na-
tional Science Foundation Grant No,
GK-1843X. This financial support is
gratefully acknowledged.
REFERENCES
1. ACI Committee 318, "Building
Code Requirements for Reinforced
Concrete (ACI 318-71)," Ameri-
can Concrete Institute, Detroit,
1971, 144 pp.
2. McGee, W. D. and Zia, P., "Pre-
stressed Concrete Members under
Torsion, Shear and Bending," Re-
search Report, Department of Civil
Engineering, North Carolina State
University, Raleigh, N.C., July
1973, 212 pp. (Also available as a
PhD. dissertation by W. D. McGee,
University Microfilm, Ann Arbor,
Michigan).
3. ACI Committee 318, "Commentary
on Building Code Requirements for
Reinforced Concrete (ACI 318-
71)," American Concrete Institute,
Detroit, 1971, 96 pp.
4. Hsu, T. T. C., "Torsion of Structur-
al Concrete—Uniformly Prestressed
Rectangular Members without
Web Reinforcement," PCI JOUR-
NAL, Vol. 13, No. 2, April 1968,
pp. 34-44.
5. Wyss, A. N., Garland, J. B., and
Mattock, A. H., "A Study of the
Behavior of I-Section Prestressed
Concrete Girders Subject to Tor-
sion," Structures and Mechanics
Report No. SM 69-1, University of
Washington, Seattle, Washington,
1969.
6. Zia, P., "Torsional Strength of Pre-
stressed Concrete Members," ACI
 Journal, Vol. 32, No. 10, April
1961, pp. 1337-1359.
7. Private communication to ACI
Committee 438, Torsion, from
Thomas T. C. Hsu, October 17,
1967.
8. Swamy, N., "The Behavior and Ul-
timate Strength of Prestressed Con-
crete Hollow Beams Under Com-
bined Bending and Torsion," Mag-
azine of Concrete Research (Lon-
don), Vol. 14, No. 40, March 1962,
pp. 13-24.
9. Johnston, D. W., and Zia, P., "Hol-
low Prestressed Concrete Beams
under Combined Torsion, Bending
and Shear," Research Report, De-
partment of Civil Engineering,
North Carolina State University,
Raleigh, N.C., December 1971, pp.
178. (Also available as a PhD. dis-
sertation by D. W. Johnston, Uni-
versity Microfilm, Ann Arbor,
Michigan)
10. Chander, H., Kemp, E. L., and
Wilhelm, W. J., "Behavior of Pre-
stressed Concrete Rectangular
Members Subjected to Pure Tor-
sion," Civil Engineering Studies
Report No. 2007, West Virginia
University, Morgantown, West Vir-
ginia, 1970.
11. Mukherjee, P., and Kemp, E. L.,
"Ultimate Torsional Strength of
Plain, Prestressed and Reinforced
Concrete Members of Rectangular
Cross Section," Civil Engineering
Studies Report No. 2003, West Vir-
ginia University, Morgantown,
West Virginia, 1967.
12. Mukherjee, P., and Warwaruk, J.,
"Prestressed Concrete Beams with
Web Reinforcement under Com-
bined Loading," Structural Engi-
neering Research Report No. 24,
Department of Civil Engineering,
University of Alberta, Edmonton,
Alberta, Canada, 1970.
13. Okada, K., Nishibayashi, S., and
64
Abe, T., "Experimental Studies on
the Strength of Rectangular Rein-
forced and Prestressed Concrete
Beams Under Combined Flexure
and Torsion," Transactions, Japa-
nese Society of Civil Engineers,
No. 131, July 1966, pp. 39-51.
14. Superfesky, M. J., "Torsional Be-
havior of Prestressed Rectangular
Concrete Beams," MS thesis, De-
partment of Civil Engineering,
West Virginia University, Morgan-
town, West Virginia, 1968.
15. GangaRao, H. V. S., and Zia, P.,
"Rectangular Prestressed Beams in
Torsion and Bending," Journal,
Structural Division, ASCE, Vol. 99,
No. ST1, January 1973, pp. 183-
198.
16. Jacobsen, E. B., "Torsion, Bending
and Shear in Rectangular Pre-
stressed Concrete Beams," MS the-
sis, Department of Civil Engineer-
ing, University of Alberta, Edmon-
ton, Alberta, Canada, 1970.
17. Woodhead, H. R. and McMullen,
A. E., "A Study of Prestressed Con-
crete under Combined Loading,"
Research Report No. CE 72-43,
University of Calgary, Calgary,
Canada, 1972. See also PCI Jour-
nal, Vol. 18, No. 5, September-Oc-
tober 1973, pp. 85-100.
18. Notes on ACI 318-71 with Design
Applications, Portland Cement As-
sociation, Skokie, Illinois, 1972, p.
12-17.
19. PCI Design Handbook, Prestressed
Concrete Institute, Chicago, Illi-
nois, 1971, p. 3 -26, and p. 3 -47.
NOTATION
A l = area of longitudinal torsion re-
inforcement
At = cross-sectional area of one leg of
stirrup
V,, = ultimate shear force
 To = contribution by concrete to tor-
sional strength of non-pre-
stressed beam
T o ' = contribution by concrete to tor-
sional strength of prestressed
beam
Tu = ultimate torque
= torsional strength of plain con-
crete beam
= concrete cylinder strength
fly = yield strength of longitudinal re-
inforcement
f, = modulus of rupture of concrete
f $y = yield strength of stirrup
h = wall thickness of box section
m = volume ratio of longitudinal re-
inforcement to web reinforce-
ment
s = spacing of stirrup
v, = permissible flexural shear stress
under combined torsion and
flexural shear
v,' = permissible flexural shear stress
without torsion as defined by
ACI Eq. (11-11) or (11-12)
v, = nominal flexural shear stress
x = shorter side of rectangular sec-
tion
xl = shorter leg of closed stirrup
y = longer side of rectangular sec-
tion
yl = long leg of closed stirrup
a = torsion constant, see Eq. (1)
fl = reinforcement coefficient, see
Eq. (6)
a- = average prestress on section
,r, = permissible shear stress for tor-
sion under combined torsion and
flexural shear
= permissible shear stress for tor-
sion without flexural shear
,r,, = nominal torsional shear stress

Discussion of this paper is invited.

Please forward your discussion to PCI Headquarters
by August 1, 1974, to permit publication in the
September-October 1974 PCI JOURNAL

PCI Journal/March-April 1974 65